基于图论和特征的图像分割

Image Segmentation Based on Efficient Graph and Feature

传统的基于图论的特征分割方法,使用图的表示方法,定义一个量去测量两个邻域间是否有边界,然后产生基于此量的分割算法,尽管此算法是贪心的,但它满足全局要求。算法是线性的,在实际中运行速度很快。在此基础上,提出用特征去处理所有的这些分支,计算分支之间的欧式距离,解决相同的非邻域分支无法正确处理的问题。

Defines a predicate for measuring the evidence for a boundary between two regions using a traditional graph-based representation of the image. Develops an efficient segmentation algorithm based on this predicate, and shows that although this algorithm makes greedy decisions, it produces segmentations that satisfy global properties. The algorithm runs in time nearly linear in the number of graph edges and is also fast in practice. Extracts feature to deal with all of the components, then calculates the distance of the components,which can solve the similar components which is not neighboring can not be classified.